A network of parametrically driven silicon nitride mechanical membranes
Luis Mestre, Suyash Singh, Gabriel Margiani, Letizia Catalini, Alexander Eichler, Vincent Dumont · 2025
Read A network of parametrically driven silicon nitride mechanical membranes by Luis Mestre, Suyash Singh, Gabriel Margiani, Letizia Catalini, Alexander Eichler, Vincent Dumont (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-00850.txt.
Introduction
Luis Mestre, Suyash Singh, Gabriel Margiani, Letizia Catalini, Alexander Eichler, and Vincent Dumont build a network of vibrating membranes that can be driven and tuned one by one, in a 2025 paper. Linked vibrating parts are a promising base for analog computing and for mimicking complex systems, but they are hard to make. A useful network needs parts that ring cleanly, tune on their own, and couple strongly to each other.
The team's parts are thin silicon nitride films coated with metal, all sharing one base, which links their motion. Electrodes steer each film on its own. The authors show they can tune each film's pitch. They can also push it with a rhythm that feeds its swing, and shift the behavior of linked films where their pitches nearly meet. They present the setup as a testbed for vibrations, sudden shifts of state, and analog computing.
Catalog record
| CREATOR | Luis Mestre, Suyash Singh, Gabriel Margiani, Letizia Catalini, Alexander Eichler, Vincent Dumont |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 46 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-00850
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A network of parametrically driven silicon nitride mechanical membranes Preprint: APS/123-QED Luis Mestre, Suyash Singh, Gabriel Margiani, Letizia Catalini, Alexander Eichler, Vincent Dumont Abstract Networks of nonlinear resonators offer a promising platform for analog computing and the emulation of complex systems. However, realizing such networks remains challenging, as it requires resonators with high quality factors, individual frequency tunability, and strong inter-resonator coupling. In this work, we present a system that meets all these criteria. Our system is based on metallized silicon nitride membranes that are coupled via their common substrate and controlled capacitively via electrodes. We demonstrate individual frequency tuning and strong parametric driving of each membrane. Notably, we tune membrane frequencies through avoided crossings and demonstrate tunability of the coupled membrane’s parametric response. This platform provides a scalable and controllable setting for exploring collective phenomena, dynamical phase transitions, nonlinear topology, and analog computing. Introduction. — Networks of coupled oscillators can emulate complex dynamical systems, offering a physical platform to solve computationally hard problems and to explore emergent phenomena in nonequilibrium physics Strogatz 2001 ; De Domenico 2023 ; Dorogovtsev et al. 2008 . For instance, they can realize Hopfield networks Hopfield 1982 used for novel forms of computing, such as Ising machines Mohseni et al. 2022 . These networks also provide a testbed for studying collective phenomena such as synchronization Arenas et al. 2008 , cascade and critical transitions Scheffer et al. 2009 ; Scheffer et al. 2012 ; Liu et al. 2024 ; Grziwotz et al. 2023 ; Harris et al. 2024 ; Artime et al. 2024 , and dynamical phases, e.g. Chimera states Abrams and Strogatz 2004 ; Matheny et al. 2019 . Kerr parametric oscillators (KPOs) are a particularly promising building block for analog computation. A KPO can be realized by modulating the spring constant of a nonlinear resonator at approximately twice its resonance frequency Lifshitz 2009 ; Bachtold et al. 2022 ; Dykman 2012 ; Eichler and Zilberberg 2023 . Above a certain modulation threshold, the system transitions into one of two stable oscillation ‘phase states’, each with equal amplitude but opposite phase. These states can be mapped to the classical states of a spin–1/2 system (up and down), and can be used as binary logic elements Mahboob and Yamaguchi 2008 ; Wilson et al. 2010 ; Puri et al. 2017 ; Mahboob et al. 2016 ; Nigg et al. 2017 ; Nosan et al. 2019 ; Frimmer et al. 2019 ; Grimm et al. 2020 ; Wang et al. 2019 ; Puri et al. 2019 ; Miller et al. 2019 ; Yamaji et al. 2023 ; Frattini et al. 2024 ; Álvarez et al. 2024 ; Han et al. 2024 ; Margiani et al. 2025 for a variety of applications Xue et al. 2025 ; Szedlak et al. 2014 ; Mendoza and Haas 2022 . Additionally, these coupled bimodal and out-of-equilibrium resonators allow for studying complex dynamical phenomena in a highly controllable and stable platform. Figure 1: Coupled silicon nitride membranes. (a) Four silicon nitride mechanical membranes (blue) are coupled through the silicon substrate (gray) to their neighbors with coupling strength J_{ij} . (b) The membranes are metallized and capacitively coupled to electrodes (gold) at a distance d_{0}\approx$13\text{\,}\mathrm{\SIUnitSymbolMicro m}$ below the membranes. We send 1550-nm laser light to a single membrane for interferometric readout. The laser is focused on the membrane with a GRIN lens, and the light reflected from the membrane is picked up by a circulator and collected by a photodiode (PD). While early theory proposals Goto 2016 ; Dykman et al. 2018 ; Puri et al. 2019 and demonstrations Margiani et al. 2023 are based on weakly coupled KPOs, much richer physics is found when the coupling strength is on the order of the damping rate Heugel et al. 2022 ; Heugel et al. 2023a ; Álvarez et al. 2024 ; Ameye et al. 2025 ; Margiani et al. 2025 . There, the interplay between nonlinearity and coupling gives rise to doubly unstable ‘ghost states’ Heugel et al. 2023a and to additional states with mixed symmetry Heugel et al. 2022 ; Margiani et al. 2025 . Strong coupling is also essential for enabling coherent transitions between states in quantum annealing Goto et al. 2018 . Aside from strong coupling, detuning between the constituent oscillators leads to an asymmetric Ising model Han et al. 2024 , which is central for modeling neural networks and other nonreciprocal systems Szedlak et al. 2014 ; Fruchart et al. 2021 . Combining strong coupling and large in-situ tuning in a scalable experimental platform would allow access to a range of important applications, including rapid phase logic operations Nosan et al. 2019 ; Frimmer et al. 2019 and programmable neural networks Han et al. 2024 . In reality, however, fulfilling all of these criteria in an experiment is difficult. In this work, we present an electromechanical platform based on coupled silicon nitride membrane resonators on a single chip. Each resonator can be individually tuned in frequency by a few \text{\,}\mathrm{kHz} , features a quality factor up to Q\sim 10^{4} , and exhibits strong nearest-neighbor coupling, with the normal mode splitting comparable or larger than the membranes mechanical linewidth. The experimental platform is scalable, as many membranes can be fabricated on a single silicon substrate. We read out multiple mechanical resonators using a single laser interferometer, significantly reducing the experimental complexity. With this setup, we demonstrate a continuous crossover between a far detuned ‘resonator regime’, with each mode comprising mostly one bare membrane oscillating, to a ‘normal-mode regime’ where we observe strongly hybridized KPOs involving both membranes. The system can thus access various scenarios that have hitherto only been studied separately. In the future, this platform will allow us investigating many-body physics and computational architectures beyond the degenerate Ising system. Device and readout. — Our device consists of a single silicon chip containing four low-stress silicon nitride membranes (Norcada NXVA-2250-5045A), see Fig. 1 (a). The 50\text{\,}\mathrm{nm} -thick membranes have lateral dimensions of $450\text{\,}\mathrm{\SIUnitSymbolMicro m}$\times$450\text{\,}\mathrm{\SIUnitSymbolMicro m}$ and are positioned 50\text{\,}\mathrm{\SIUnitSymbolMicro m} apart from each other in a two-by-two grid. For electrostatic tuning and actuation, we coat the membranes with Ti ( 10\text{\,}\mathrm{nm} ) and Au ( 30\text{\,}\mathrm{nm} ) using optical lithography and e-beam evaporation, see Fig. 1 (b) and Appendix A . The centers of the membranes, where we optically read out the membrane oscillations, are left uncoated to minimize heating due to light absorption. The chip is positioned and clamped on top of spacers on a printed circuit board (PCB), setting a distance d_{0}\approx$13\text{\,}\mathrm{\SIUnitSymbolMicro}$ between the membrane surface and the PCB’s electrodes. The membrane chip and PCB are placed in a vacuum chamber (pressure \sim${10}^{-7}\text{\,}\mathrm{mbar}$ ) at room temperature. Figure 2: Characterization of single membrane resonator. (a) Measured (dots) and fitted (solide lines) mechanical response amplitude (blue) and phase (grey) as a function of the driving frequency \omega . Fitting to the resonance curve x_{1}(\omega)\propto(\omega_{1}^{2}-\Omega^{2}-i\omega\Gamma_{1})^{-1}U_{1}(\omega) yields \Omega_{1}/2\pi\approx$202.67\text{\,}\mathrm{kHz}$ and damping \Gamma_{1}/2\pi=$12.59\pm 8\text{\,}\mathrm{Hz}$ . (b) Membrane frequency \Omega_{1}+\delta\Omega_{1} as a function of applied DC voltage \bar{U}_{1} . From a quadratic fit, we extract the membrane-electrode distance d_{0}\approx$13\text{\,}\mathrm{\SIUnitSymbolMicro m}$ , see Appendix C . (c) Amplitude (orange) and phase (grey) measured at \omega in response to parametric driving at 2\omega with